Sunday, April 4, 2010

More GHCN results.

1 comment:

  1. Nice work Nick... something else you may want to do: add uncertainties to your series and trend fits. Part of doing the trend fit is remembering that your uncertainty generally is increasing as your number of stations drop (especially before 1950), i.e., so really one also needs to weight by year, if you fit to an interval before that.

    Of course, this is already done in CRUTEM3, so if you haven't seen it, here is a plot including 95% uncertainty bounds